"Flat" is a misnomer.
If we use it loosely to mean "linear" (i.e. non-distorting) sound reproduction device, then I agree with you completely.
That's one of the goals that audio reproduction engineers aim for: the truth in audio reproduction. Reproduce the recorded signally with as little of distortion as possible (in terms of signal theory).
However, even in this scenario there is a known (but unexplainable) discrepancy between linear components and the actual experience of listening to them.
Not all "flat" devices (using the term loosely here, as described above) sound equal. Also, some of them sound worse than "non-flat" devices. This cannot be explained fully and without any room for error using known acoustic/psychoacoustic science of today. Or if it can be done, I'd very much like to receive a reference to a scientific literature that explains this.
Now, the second meaning for the term "flat" refers to it's spectral output: namely that it measure absolutely flat in it's amplitude response (as a function of frequency), when measured acoustically (not just electrically).
However, and this is a big however, headphones do not measure flat when they are "ideal".
Why?
Because:
1) The point at which headphones are measured (concha, entrance to ear canal) is not "flat" in the human ear. If you don't wear headphones, but just wear microphones at concha level and record sounds from absolutely flat speakers (in a diffuse field or anechoic chamber), then the sound at your concha level is not flat anymore. The graph will NOT look like a straight line.
This is due to what is overall called a head related transfer function, which takes into account the torso, head (including hair) and outer ear reflections and masking.
This is "natural" at it's best. You can't get more natural than that frequency response curve at your concha level. The sound gets even more distorted as it enters the ear canal (and then ot middle ear and inner ear), but for the sake of simplification we'll leave it out of the discussion now.
That response (and related HRTF) is also individual to every single living person.
No two persons hears one sound exactly the same, although there is a high degree of agreement on what the sound sounds like (probably due to neural matching at higher cognitive levels of auditory processing).
2) Second reason why ideal headphones do not measure flat is because with headphones on, one is measuring a head+headphone coupled system. Not just the headphone.
If one designs headphones so that they are ideal in their frequency response when not worn, they will not be ideal anymore when worn, due to the reflective chamber that the head+headphoen produces. A good headphone designer takes this into account and designs the headphone to sound best (not flat, but ideal in terms of generic/averaged HRTFs) when worn.
So, in summary:
- an ideal (in terms of best known acoustic measurements correlated with listening test) headphone graph when measured from entrance to ear canal (on a person's head, i.e. worn) is not a straight line (i.e. not "flat" in that sense of the word).
- designing/measuring headphones to be flat at that position will make them sound unnatural. Yes, it's possible to make them more "reavealing" this way due to the masking effect of human hearing.
However making them more revealing will reveal things on a recording that a human listener would NOT be able to hear in a normal acoustic setting. This is what I sometimes call personally the desire of the Hyperreal or desire to augment senses beyond what they are naturally capable of (like x-ray acoustics). Also, the process will simultaneously mask other frequencies, that were earlier audible before. That is the nature of human auditory system and cannot be mechanically avoided (maybe with some neural implants in the future, but even then it begs to ask the question: "why? for what use?").
These two things are just some of the small things why (IMHO) making snap judgements about flat frequency response curves can be very misleading indeed.
In headphone measurements, flat is not "ideal" (in objective scientific terms).
It can however be most pleasing to some people. We all like many different things, after all.
regards,
halcyon
PS Thanks for the welcome back greetings gerG! don't have that much time for headfi these days, but I try to pop-in every now and then to see your new measurements